Journal of Cleaner Production 6 (1998) 277–288

Screening life cycle assessment (LCA) of tomato ketchup: a case

study
Karin Andersson *, Thomas Ohlsson, Pär Olsson
SIK, The Swedish Institute for Food and Biotechnology, P.O. Box 5401, SE-402 29 Göteborg, Sweden

Abstract

A screening life cycle assessment (LCA) of tomato ketchup has been carried out. The purpose was to identify ‘hot-spots’, that
is parts of the life-cycle that are important to the total environmental impact. The system investigated includes agricultural pro-
duction, industrial refining, packaging, transportation, consumption and waste management. Energy use and emissions were quant-
ified and some of the potential environmental effects assessed. Packaging and food processing were found to be hot-spots for many,
but not all, of the impact categories investigated. For primary energy use, the storage time in a refrigerator (household phase) was
found to be a critical parameter.  1998 Elsevier Science Ltd. All rights reserved.

Keywords: Life cycle assessment; LCA; Foods; Tomato products; Ketchup

1. Introduction to cover the entire life cycle of a food product; for

The current systems for food production require large
inputs of resources and cause several negative environ-
mental effects. The systems are optimised to satisfy
economic demands and the nutritional needs of a rapidly
growing world population. Environmental issues, how-
ever, have not been central.
There are many difficulties in conducting life cycle
studies of food products. Ideally, a complete study
should include agricultural production, industrial
refining, storage and distribution, packaging, consump-
tion and waste management, all of which together com-
prise a large and complex system. The lack of public
databases hinders collection of suitable data. Another
difficulty is that life cycle studies involve many scientific
disciplines. In the 1970s and early 1980s, the use of
energy in food production systems was widely studied
[1,2]. Most food life cycle studies carried out so far treat
either agricultural production or industrial refining; for
example, the cultivation of tomatoes has been studied
[3]. There have been only a few studies that attempted
example, a screening life cycle inventory (LCI) of rye
bread and ham with emphasis on establishing the energy
and material flows [4].
The aim of the study presented here was to carry out
a screening life cycle assessment (LCA) to learn more
about the options and limitations of applying the method
to food production systems. Tomato ketchup was chosen
as a suitable product. As the work was done in close co-
operation with a Swedish producer of tomato ketchup
and an Italian producer of tomato paste, it was possible
to obtain a large amount of site-specific inventory data.
The impact assessment made includes the following
environmental effects: global warming, ozone depletion,
acidification, eutrophication, photo-oxidant formation,
human toxicity and ecotoxicity. Intermediate inventory
results have already been reported [5]. A more thorough
description of the model system, the assumptions made
and the data used, can be found in the comprehensive
report (in preparation).
2. Method

2.1. Goal definition

* Corresponding author. Current address: CIT Ekologik, Chalmers
Teknikpark, SE-412 88 Göteborg, Sweden. Tel.: ⫹ 46-31-772-4334; The main goal of the case study, part of a research
Fax: ⫹ 46-31-82-7421; E-mail: karin.andersson@cit.chalmers.se project funded by the Swedish Waste Research Council,

0959-6526/98/$19.00  1998 Elsevier Science Ltd. All rights reserved.

PII: S 0 9 5 9 - 6 5 2 6 ( 9 8 ) 0 0 0 2 7 - 4
278 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288
was to identify key issues associated with the life cycle
of tomato ketchup, such as (1) the steps of the life cycle
which give rise to the most significant environmental
input and output flows, that is hot spots, and (2) major
gaps in the data available. The comprehensive report (in
preparation) includes, in addition to the screening LCA:
a comparison of the current packaging system for
ketchup and an alternative one, and an improvement
assessment of a selected part of the life cycle.
2.2. The product and the system investigated
The product studied is one of the most common
brands of tomato ketchup sold in Sweden; it is marketed
in 1 kg red plastic bottles. The complete system investi-
gated is shown in Fig. 1. The packaging systems for tom-
ato paste and ketchup are shown in Figs. 2 and 3,
respectively. The life cycle can be described briefly as
follows. Tomatoes are grown and processed into tomato
paste in the Mediterranean countries; then the tomato
paste is transported to Sweden and processed (together
with other ingredients and water) into ketchup. There-
after, the ketchup is packaged, delivered to retailers and,
finally, consumed. The tomato paste is packed in aseptic
bags which are placed in steel barrels. Each bag contains
200 l of tomato paste. The plastic bottle used for ketchup
is made of polypropylene (PP) and is blow-moulded. It
consists of five layers: an inner wall of PP; adhesive; a
barrier layer of ethylenevinylalcohol (EVOH); adhesive;
and an outer wall of PP.
The model system was divided into six subsystems.
For the packaging and household subsystems, alternative
scenarios were analysed. Table 1 shows a summary of
the subsystems, the processes they include and the scen-
arios investigated. For the packaging subsystem, the
waste management scenarios investigated are further
defined in Table 2.
2.3. The functional unit
The functional unit (FU) is defined as 1000 kg of tom-
ato ketchup consumed, assuming a 5% loss in the house-
hold phase. We prepared a questionnaire which was
answered by 30 persons and collected their ketchup
bottles at the point of disposal. This very limited survey
indicated: (1) that the household scenarios are realistic
as to storage time; and (2) that the losses vary signifi-
cantly, that is values from 0.5% to 26% were recorded.
The 5% loss assumed was validated as a reasonable esti-
mate; other losses can easily be simulated with the scen-
ario technique.
2.4. The inventory analysis and data collection
For the inventory analysis, a summary of the pro-
cesses included, the data sources and the principles of
allocation applied are shown in Table 3. Our ambition
was to use site-specific inventory data, whenever poss-
ible. To collect this data and other information, we used
our own questionnaire, interviews and environmental
reports. The Swedish producer of ketchup has several
suppliers of tomato paste in Italy, Spain and Portugal;
the inventory data for tomato paste were collected from
one of the suppliers in Italy. The data for cultivation of
tomatoes were collected from one of the farms supplying
the Italian tomato paste plant. Most of the inventory data
were collected in 1993 and 1994.
2.5. The system boundaries
Procedures and results from the inventory analysis
have already been presented [5]. Since then, the system
investigated has been expanded to include: production
of electricity; cultivation of sugar beets; production of
raw sugar; treatment of the waste water from production
of both the sugar solution and the ketchup; shopping;
and the household phase. Thermal energy was accounted
for as the amount used for combustion. Whenever emis-
sion factors were used, emissions from the fuel extrac-
tion (precombustion emissions) were included. For elec-
tricity from a grid, average country-specific data for
electricity production were used when the geographic
location of the process was known. Otherwise average
figures for European electricity were used (see Table 3).
These figures are based on data for the ‘Union for the
Connection of Production and Transportation of Elec-
tricity, UCPTE 88’ as presented by BUWAL [6].
To compare the environmental impacts of the two
waste management scenarios for the packaging materials
(see Tables 1 and 2), system expansion with a marginal
substitute, oil, was applied [9,13]. Thus, for waste incin-
eration, the energy recovered was subtracted from the
scenario’s total energy use. The energy recovered was
also assumed to mean a reduced need of oil for heating
purposes. The emissions from scenarios including waste
incineration were therefore adjusted: the emissions that
would have resulted, if oil had been used to generate the
amount of energy recovered, were subtracted from the
total emissions of the scenario.
The treatment of waste water, at the municipal plant,
was included for the production of ketchup and sugar
solution; the other food processing plants have their own
waste water treatment. Since the municipal plant treating
the effluent from the ketchup production receives 90%
of its load from this particular food industry, site-specific
data from this waste water treatment plant were used.
For treatment of the effluent from sugar solution pro-
duction, general data on efficiencies and energy use for
an assumed waste water treatment plant with mechan-
ical, biological and chemical treatment were used [12].
This type of plant is the most common in Sweden.
Due to data gaps, the following steps were left outside
 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288 279

Fig. 1. Principal flow chart of the life cycle of the Swedish tomato ketchup.

the system boundaries: the production of capital goods
(machinery and buildings); the production of citric acid;
the wholesale dealer; transportation from the wholesaler
to the retailer; and the retailer. Likewise, for the ketchup
bottles, the production of adhesive, EVOH, pigment, lab-
els, glue and ink were omitted due to lack of accessible
data. The aseptic bags used for the tomato paste contain
7% polyethyleneterephthalate (PET) and 0.03% alu-
minium; these materials were omitted, due to the small
amounts in which they occur. For the household phase,
leakage of refrigerants was left outside the system
boundaries.
In the cultivation steps, the assimilation of CO2 by
the crops was not taken into consideration; neither was
leakage of nutrients and gaseous emissions such as
ammonia and nitrous oxide from the fields. Models for
280 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288

ing factors for many of the active substances involved

are unavailable.

2.6. Methodological choices, assumptions and

simplifications

For the production of tomato paste and ketchup, allo-

Fig. 2. The tomato paste packaging system investigated.
doing this need to be worked out. For pesticides, a quan-
titative inventory and impact assessment were found to
be beyond the scope of this study. The types and
amounts of pesticides applied are closely related to the
weather, which means that inventory data need to be col-
lected for longer time periods. In addition, models for
leakage and degradation mechanisms as well as weight-
cation was made by weight. The results obtained were
validated in the following ways.
쐌 At the tomato paste plant, mass allocation yields the
following requirements per tonne of product: 5.9 GJ
thermal and 0.38 GJ electrical energy. As to thermal
energy use, the specific production line is dominated
by the evaporation and sterilisation steps for which
data on the use of steam and electricity were collected
and requirements of 5.1 GJ thermal and 0.18 GJ elec-
trical energy per tonne tomato paste were calculated.

Fig. 3. The ketchup packaging system investigated. LDPE is short for low density polyethylene. The dotted lines around EVOH indicate that the
production of EVOH was not included within the system boundaries.

Table 1
The ketchup production subsystems, the processes included and the scenarios investigated

Subsystem Processes included Scenarios

Agriculture Cultivation of tomatoes and sugar beets. Production of inputs to the

cultivation steps
Food processing Production of tomato paste, raw sugar, sugar solution, vinegar, spice
emulsion, salt and ketchup
Packaging Production and transportation processes included in the packaging Waste management: (1) Landfill, (2) Material
systems for tomato paste and ketchup recycling and/or incineration with energy recovery
Transportation All transportation processes except for the transports included in the
packaging subsystem
Shopping Transportation from retailer to household
Household Storage of ketchup bottle in refrigerator Storage time: (A) One month, (B) One year
 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288 281

Table 2
The tomato paste and ketchup packaging systems and the different waste management scenarios

The tomato paste packaging system The ketchup packaging system

Scenario 1 Steel barrels, plastic materials and wood pallets: to landfill Plastic materials: to landfill. Corrugated cardboard: 80% to
recycling and 20% to landfill. Wood pallets: reused 100 times,
then to landfill
Scenario 2a Steel barrels: 70% to recycling and 30% to landfill. PP: 80% to LDPE: to incineration. PP: 80% to incineration and 20% to
incineration and 20% to landfill. LDPE and wood pallets: to landfill. Corrugated cardboard: 80% to recycling and 20% to
incineration incineration. Wood pallets: reused 100 times, then to
incineration

a
Incineration means that the inherent energy is recovered.

Table 3
The processes included, the sources of data and the principles of allocation used for the inventory analysis

Process Type or source of data Source of Principle of allocation used

emission factors

Fertiliser N [4] [6]a cb

Fertilisers P and K [7]a [4]b [6]a cb
Pesticides [4] [6]
Lubricating oils [7] [6]
Seeds [7] [6]
Tomatoes Site-specific estimates, Italy [6]
Sugar beets [4] and site-specific estimates, Sweden [8]
Tomato paste Site-specific, Italy [6] By weight, for the main products of the plant
Raw sugar Site-specific, Sweden c
Sugar solution Site-specific, Sweden [9]
Vinegar Site-specific, Sweden
Spice emulsion Site-specific, Sweden
Salt [4] [9]
Tomato ketchup Site-specific, Sweden c For emissions caused by use of thermal energy, by
weight for the main products of the plant. For water
emissions, by flows of water through the different
production lines
Packaging system for [6]: PP, [9]: steel, LDPE and wood
tomato paste
Packaging system for [6]: PP, [9]: LDPE, corrugated cardboard
ketchup and wood. Site-specific: blow-moulding and
red master-batch
Transportation Site-specific: type of vehicle and distance [8]
Shopping Own estimates [10]
Household phase Own estimates
Electricity production Average, country-specific. Average, Europe c, [11]
Waste management Packaging materials, as specified above.
Waste water: site-specific and [12]

a
Tomatoes.
b
Sugar beets.
c: The software ‘LCA Inventory Tool’ (LCAiT).

Thus, the results obtained by mass allocation appear
reasonable. Besides tomato paste, canned peeled
whole tomatoes and diced tomatoes are produced.
Literature data were used to check the stability of the
results; for canned fruits and vegetables, requirements
of 5.2 GJ thermal and 0.20 GJ electrical energy per
tonne have been reported [2].
쐌 Co-products in the ketchup plant are salad dressing,
cooking oil, jam, mayonnaise and horse-radish; how-
ever, the ketchup dominates by mass: 66% of the total
production. Mass allocation yields the requirement of
1.7 GJ thermal energy per tonne of product. (The
requirement of 0.38 GJ electricity was recorded for
the ketchup production line.) According to literature
data, the energy requirements of the product groups
‘pickles, sauces and salad dressings’ and ‘cooking
oils’ are of similar magnitude (ketchup can be
regarded as a sauce). For pickles, sauces and salad
dressings, the energy requirement 2.6 GJ thermal plus
0.48 GJ electrical energy per tonne has been reported;
282 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288
for cooking oils, the corresponding figures are 3.0 GJ
thermal plus 0.24 GJ electrical energy [2]. Since jam
is heated in a process similar to that used for ketchup,
the thermal energy requirement per mass unit should
be comparable to the one for ketchup. Mayonnaise
and horse-radish are produced in relatively small vol-
umes.
To enable including the transportation of ketchup
between the retailer and the consumer, many assump-
tions had to be made. The basic parameters for calculat-
ing the environmental loads caused by shopping are:
쐌 the proportion of trips made by car, assumed to be
55%;
쐌 the distance driven, assumed to be 2.5 km each
way; and
쐌 the amount of groceries bought, assumed to be 10 kg.
The environmental loads were allocated by weight for
the products bought.
To include the loss occurring in the household phase,
all of the results from the inventory analysis were multi-
plied by 1.05. This was done to link the environmental
loads caused by the losses to their actual geographic
location. For the rest of the system, the calculations were
carried out as if there were no losses. For example, all
of the tomatoes harvested were assumed to be used in
the production of tomato paste. All of the tomato paste
produced was assumed to be used in the production of
tomato ketchup, and so on.
For the cultivation steps, we have assumed: that only
fertilisers (no manure) are used; and that agricultural
land is contaminated with 660 mg zinc, 100 mg arsenic
and 60 mg cadmium per kg phosphorous applied as ferti-
liser [4].
2.7. The impact assessment
In the classification and characterisation done we have
followed the Nordic Guidelines [12]. The contributions
to the following impact categories were assessed.
쐌 Global warming was obtained, for direct greenhouse
gases, by using Global Warming Potentials (GWPs)
of the time-horizons 20, 100 and 500 years [14].
Indirect greenhouse gases were included to check the
influence on the results [15,16].
쐌 Depletion of stratospheric ozone was worked out by
using the inventory results for methane, nitrous oxide,
carbon monoxide and non-methane hydrocarbons,
since these substances contribute directly or indirectly
to the effect. No substances for which Ozone
Depletion Potentials (ODPs) are available were ident-
ified in the inventory analysis.
쐌 Acidification was assessed by using the ‘protons
released approach’ with its minimum and maximum
scenarios.
쐌 Eutrophication was found by means of the ‘scenario-
based approach’ which divides the category into five
subcategories. As a complement to the characteris-
ation results, the inventory parameter BOD
(Biological Oxygen Demand) was taken into con-
sideration; for some processes BOD is known but not
COD (Chemical Oxygen Demand), and vice versa.
쐌 Photo-oxidant formation was obtained by using the
concept of Photochemical Ozone Creation Potentials
(POCPs) [15,17,18] and inventory results were used
for substances that lack available weighting factors.
쐌 Human toxicity was assessed by using the CML pro-
visional and Tellus methods; ecotoxicity was handled
with the CML provisional method [18,19]. In
addition, for radioactive waste and radon, the inven-
tory results were taken into consideration.
3. Results
The use of primary energy and potential contributions
to global warming, ozone depletion, acidification,
eutrophication, photo-oxidant formation, human toxicity
and ecotoxicity, at subsystem level, are presented next.
3.1. Energy use
The use of primary energy and the energy sources are
shown in Fig. 4. The scenario for the household subsys-
tem (time for storage in the refrigerator) is critical. The
figure used for storage in the refrigerator is 4.73 Wh per
litre and day [4]. Primary energy use in this subsystem
varies between approximately 10% and 50% of the total.
It is clear that the energy requirements of the food pro-
cessing and the packaging subsystems are also
important. In food processing, approximately one third
of the energy requirement is for the production of tomato
paste; one third for the other ketchup ingredients; and
one third for the ketchup itself. For the packaging subsy-
stem, the scenario is not as critical as for the household
subsystem. Note that the transportation subsystem and
the process of shopping have similar energy require-
ments. The contribution of transportation to the packag-
ing subsystem is not known, since the form of literature
data does not allow one to distinguish efficiently the
energy used for production from that used for transpor-
tation.
3.2. Global environmental effects
The characterisation results for global warming are
shown in Fig. 5. The food processing and the packaging
subsystems make large contributions to global warming
because of their high consumption of fossil fuels. The
low contribution from the household subsystem is due
 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288 283

Fig. 4. The use of primary energy in the ketchup production system.

Fig. 5. The assessed contributions to global warming.

to the assumption that only Swedish electricity is used
and the fact that the Swedish electricity model is domi-
nated by hydropower and nuclear power; for a 100-year
time frame, the contributions made by the households A
and B are approximately 2.2 and 26 g CO2-equivalents
per functional unit (FU), respectively. The contributions
made by indirect greenhouse gases are, except for the
process of shopping, relatively small; they decrease with
longer time frames.
For ozone depletion, the inventory results for methane
(CH4), nitrous oxide (N2O), carbon monoxide (CO) and
non-methane hydrocarbons (NMHC) are shown in
Table 4.
3.3. Regional environmental effects
The contributions assessed for the regional environ-
mental effects, acidification, eutrophication and photo-
oxidant formation, are shown in Tables 5–9. For acidifi-
cation, the food processing subsystem is an obvious hot-
spot (see Table 5). Since the geographic location is of
significance, it is relevant to analyse the food processing
subsystem further. Depending on the characterisation
model chosen, the sulphur dioxide (SO2) emitted in the
production of tomato paste is responsible for between
70% and 90% of the effect. This life cycle step is here
located specifically in northern Italy; the combination of
high energy use and the choice of fuel (heavy fuel oil)
is the cause of the high SO2-emission. The reason for
the negative contribution shown by the Packaging 2
scenario is an avoided emission of SO2, due to the
assumption that the energy recovered from waste incin-
eration replaces heat produced by combustion of oil. The
differences between the results of the minimum and
maximum characterisation models vary. The minimum
model excludes the acidifying potential of nitrogen com-
284 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288

Table 4
Inventory results for emissions contributing to ozone depletion in g per FU

Subsystem CH4 N2O CO NMHC

Agriculture 21 130 370 420

Food processing 620 38 62 540
Packaging 1 120 5.7 160 900
Packaging 2 120 5.7 210 860
Transportation 330 98
Shopping 29 2.2 7600 770
Household A 7.8 ⫻ 10−5 2.3 ⫻ 10−4 1.2 ⫻ 10−3 6.1 ⫻ 10−4
Household B 9.1 ⫻ 10−4 2.7 ⫻ 10−3 1.4 ⫻ 10−2 7.2 ⫻ 10−3

Table 5 water treatment, the inventory results for the parameter

Characterisation results for acidification in mol H+ per FU BOD are presented in Table 7.
Subsystem Minimum Maximum Characterisation results for photo-oxidant formation
are shown in Table 8; inventory results for nitrogen
Agriculture 8.4 38 oxides and other organic compounds are shown in Table
Food processing 94 120 9. For the characterisation results, it is important to note
Packaging 1 15 25 which substances are included in the different character-
Packaging 2 ⫺ 7.4 3.7
Transportation 13 44 isation models. While the ‘Local, Europe’ model
Shopping 1.2 9.0 includes the emissions of hydrocarbons (HC), aldehydes,
Household A 1.1 ⫻ 10−4 2.3 ⫻ 10−4 ethanol and methane, the other models include only the
Household B 1.4 ⫻ 10−3 2.7 ⫻ 10−3 emissions of ethanol and carbon monoxide. Ethanol is a
parameter that occurs only in the production of vinegar;
hence the relatively high contribution to photo-oxidant
pounds; thus, the smaller the emissions of nitrogen com- formation from the food processing subsystem. The high
pounds from a subsystem, the less difference between contribution made by the packaging subsystem, accord-
the models. ing to the ‘Local, Europe’ model, is due to the emissions
For eutrophication, the agriculture subsystem is an of HC.
obvious hot-spot, even though leakage of nutrients in the
cultivation steps was omitted (see Table 6). The rela-
tively high contribution made by agriculture to the ‘P- Table 7
limited’ subcategory is due to emissions of phosphate Inventory results for biological oxygen demand (BOD)
from the production of phosphorous fertilisers. For the Subsystem BOD
‘N to air’ subcategory (terrestrial eutrophication), the (kg per FU)
agriculture, transportation and food processing subsys-
tems are hot-spots. When emissions of nitrogen com- Agriculture 8.1 ⫻ 10−5
pounds dominate, the ‘N-limited ⫹ N to air’ and Food processing excluding waste water treatment 8.4
Food processing including waste water treatment 0.27
‘Maximum’ subcategories show similar results. As a Packaging 1 0.15
complement to the characterisation results, as well as to Packaging 2 0.15
highlight the effect of including or excluding the waste

Table 6
Characterisation results for eutrophication in kg O2 per FU

Subsystem P-limited N-limited N to air N-limited plus N to air Maximum

Agriculture 31 8.4 9.7 18 49

Food processing 0.79 1.2 6.9 8.1 8.4
Packaging 1 0.35 0.35 2.8 3.1 3.1
Packaging 2 0.35 0.35 3.1 3.5 3.5
Transportation 1.1 ⫻ 10−3 5.3 ⫻ 10−3 8.4 8.4 8.4
Shopping 1.2 ⫻ 10−3 5.9 ⫻ 10−3 2.2 2.2 2.2
Household A 3.4 ⫻ 10−5 3.4 ⫻ 10−5 3.4 ⫻ 10−5
Household B 3.9 ⫻ 10−4 3.9 ⫻ 10−4 3.9 ⫻ 10−4
 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288 285

Table 8
Characterisation results for photo-oxidant formation in g ethene-equivalents per FU

Subsystem Regional, Swedena Local, Swedena Regional, Europea Local, Europeb

Agriculture 15 13 12 160
Food processing 32 60 43 190
Packaging 1 6.6 5.9 5.3 340
Packaging 2 8.6 7.7 6.8 330
Transportation 13 12 11 37
Shopping 300 270 240 290
Household A 4.8 ⫻ 10−5 4.3 ⫻ 10−5 3.8 ⫻ 10−5 2.3 ⫻ 10−4
Household B 5.6 ⫻ 10−4 5.0 ⫻ 10−4 4.5 ⫻ 10−4 2.7 ⫻ 10−3

a
Assessed by using the POCPs of Finnveden et al. and Andersson-Sköld et al. [15,17].
b
Assessed by using the POCPs of Heijungs et al. [18].

Table 9 Table 11
Inventory results for substances (without weighting factors) contribu- Contributions to human toxicity assessed according to the Tellus
ting to photo-oxidant formation method

Subsystem NOx Other organic compounds Subsystem Carcinogens Non- Combined

(kg per FU) (g per FU) (g isophorone- carcinogens ranking
equiv. per FU) (g xylene-equiv.
Agriculture 1.1 1.2 per FU)
Food processing 1.1 3.4
Packaging 1 0.45 Agriculture 4100 23,000 28,000
Packaging 2 0.51 Food processing 8.0 8.0
Transportation 1.4 Packaging 1 4.2 4.2
Shopping 0.36 Packaging 2 4.2 4.2
Household A 5.1 ⫻ 10−6 Transportation 0.019 0.019
Household B 6.0 ⫻ 10−5 Shopping 0.021 0.021
Household A
Household B

3.4. Toxicity
The characterisation results for human toxicity and
ecotoxicity are presented in Tables 10–12. According to
both the Tellus method for human toxicity and the CML
provisional method for ecotoxicity, the agriculture sub-
system is a hot-spot even though leakage of pesticides
was not quantitatively included. The reason is the con-
tent of heavy metals in phosphorous fertilisers. The
inventory results for radioactive waste (caused by the
production of electricity) and emissions of radon (caused
by the extraction of coal) are presented in Table 13.
4. Conclusions and discussion
The most important goal of any life cycle study is, of
course, to improve and optimise the system. Based on
the study carried out, we have identified parts of the life
cycle that are critical to the total environmental impact
as well as some major gaps in the available data. The
use of energy has often been employed as an indicator
of environmental impact. The results presented illustrate
the complexity in a scientific evaluation of a product’s
environmental performance; the results of the energy

Table 10
Contributions to human toxicity assessed according to the CML provisional method in kg body weight per FU

Subsystem Air emissions Water emissions Soil emissions Total

Agriculture 1.2 0.026 1.3 ⫻ 10−3 1.2

Food processing 4.5 1.9 ⫻ 10−6 4.5
Packaging 1 0.94 1.8 ⫻ 10−5 0.94
Packaging 2 0.11 1.7 ⫻ 10−5 0.11
Transportation 1.6 3.0 ⫻ 10−7 1.6
Shopping 0.41 3.4 ⫻ 10−7 0.41
Household A 8.4 ⫻ 10−6 8.4 ⫻ 10−6
Household B 1.0 ⫻ 10−4 1.0 ⫻ 10−4
286 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288

Table 12 packaging subsystems were found to be hot-spots. The

Contributions to ecotoxicity assessed according to the CML pro-
visional method
Subsystem Water emissions Soil emissions
(m3 per FU) (kg soil per FU)
Agriculture 180 8600
Food processing 680
Packaging 1 2700
Packaging 2 2600
Transportation 58
Shopping 66
Household A
Household B
analysis do not always point in the same direction as
those of the impact assessment. The need for simulations
to facilitate environmental improvements and optimis-
ation is evident.
For many of the impact categories, the packaging and
food processing subsystems were found to be hot-spots.
For primary energy use, the length of time for storage
in a refrigerator (household phase) was found to be a
critical parameter. With a storage time of one year, the
use of primary energy in the household phase is as high
as the energy use of the packaging and food processing
subsystems together. An example of an impact category
with a different result is eutrophication; for this effect,
the agriculture subsystem is an obvious hot-spot. For the
impact categories ozone depletion and photo-oxidant
formation, it is not possible to draw any general con-
clusions. For ozone depletion, each parameter from the
inventory must be evaluated separately. However, as
long as freons from refrigerators leak, the household
phase can be expected to contribute significantly. For
photo-oxidant formation, it is necessary to compare care-
fully the results from the different characterisation mod-
els and to keep in mind the parameters they include.
According to the characterisation results, the shopping,
packaging and food processing subsystems are hot-spots.
For NOx, the transportation subsystem is a hot-spot.
For toxicity, the agriculture, food processing and
CML provisional method and the Tellus method for
human toxicity provide different but complementary
results. One of the reasons is that the CML provisional
method includes some of the more common emissions
to air (for example SO2, NOx and CO), while the Tellus
method does not. Consequently, life cycle steps with
high use of fossil energy show high contributions to
human toxicity when the CML provisional is used. Eco-
toxicity hot-spots are life cycle steps with emissions of
heavy metals, phenol or crude oil. If leakage of pestic-
ides, their intermediates and break-down products had
been quantitatively considered, the agriculture subsys-
tem would have been an even worse toxicological hot-
spot.
Although uncertainties have not been quantified, they
are expected to be relatively large. The results from the
energy analysis are more accurate than the characteris-
ation results; figures for fuel or electricity consumption
of a process, at least at plant level, are usually available
and accurate since they represent costs. The fact that cer-
tain types of emissions represent costs, for example CO2
taxation in Sweden, has lead to monitoring and a search
for methods of reduction. However, emission data are
less exact in general than figures on energy use. The
characterisation models introduce additional uncer-
tainties; however, they make it easier to interpret the
results since the parameters from the inventory are
numerous.
The most important omissions and their possible
influence on the results are summarised below.
쐌 Production of capital goods was left outside the sys-
tem boundaries. The steps most likely to be affected
by including the production of capital goods are culti-
vation and the production of tomato paste. In the culti-
vation step many different machines are used and each
farm usually has its own. Many of these machines are
used only a few times per year. Similarly, tomato
paste is produced for only a few weeks each year.
The energy requirements of capital goods used for the
cultivation of tomatoes was estimated, using literature

Table 13
Inventory results for radioactive waste and radon

Subsystem High radioactivity Medium radioactivity Low radioactivity Rn-222

(cm3 per FU) (cm3 per FU) (cm3 per FU) (Bq per FU)

Agriculture
Food processing 0.78 8.9 8.9 400
Packaging 1 0.71 8.1 8.1 180
Packaging 2 0.71 8.1 8.1 180
Transportation
Shopping
Household A 7.5 ⫻ 10−4 8.5 ⫻ 10−3 8.5 ⫻ 10−3
Household B 8.8 ⫻ 10−3 0.10 0.10
 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288
data for France, to be 180 MJ per FU including
machinery and buildings [4]. Adjusting the calcu-
lations made for the cultivation of sugar beets (site-
specific data), results in an extra requirement of 193
MJ per FU. Altogether, this corresponds to approxi-
mately 30% of the total estimated energy requirement
of the agriculture subsystem.
쐌 The wholesale and retail step was also left outside the
system boundaries. Literature data for this life cycle
step indicate that the energy requirements for other
products are not at all negligible: 1.43 MJ per kg beer
for storage at the wholesale trader in Switzerland and
1.66 MJ per kg bread in the Netherlands [20,21]. Data
on the total annual energy use of the wholesale and
retail step are available for Sweden [22], but there is a
major data gap that hinders quantification of the share
which should be allocated to ketchup.
The practical application of this study is to provide a
platform for improvement analyses. The calculation
model constructed (simple spreadsheets in Microsoft
Excel) allows for simulations by means of which alterna-
tive scenarios for each step can be calculated and evalu-
ated. The main problem encountered in our endeavour
to apply the LCA methodology to a food system was,
besides the great gaps in accessible data, how to handle
the agricultural production and the consumer phase; for
both these phases, collection of representative data is
only one difficulty. Agricultural production makes spe-
cial demands on the LCA methodology [4,23–26]. For
instance, it is difficult to determine the system boundary
between the technological system and nature; agricul-
tural production takes place in nature itself and is actu-
ally a part of the environmental system. Ideally, all of
the crops in a crop rotation system should be studied,
since a crop may be influenced by the previous crops;
the environmental loads should then be allocated
between the different crops. An allocation problem is
how to handle common agricultural co-products such as
straw and animal manure. Models to estimate the leak-
age of nutrients and pesticides in cultivation, for differ-
ent soils, climate and crops, are needed in LCAs of food
products; the models for characterisation of human tox-
icity and ecotoxicity also need further development. In
particular, the energy analysis indicates that the house-
hold phase and the behaviour of the consumer in con-
junction with shopping (car use, distance and amount
bought) may be very important. In addition, the 5% loss
in the household phase is equivalent to 5% of the total
environmental impact. Accordingly, emphasis on the
household phase is recommended in future studies.
In the food processing industries, the data required for
life cycle studies are seldom available for the specific
production line; thus, allocation or measurements must
be conducted. Measurements are very valuable for vali-
dation of simple allocation principles for common pro-
287
ducts and processes in the food industry. For food sys-
tems, the handling of waste water deserves more
attention. Waste water treatment requires energy and
chemicals; this produces emissions. It would be relevant
to simulate alternative scenarios for the waste water
handling. In the right place, waste water from the food
industry would be a source of nutrients or a resource for
production of biogas.
The case study reported here is one of the first LCAs
of a whole food system. In spite of its limitations, it is
a rather complete study and the collection of site-specific
data contributes to the high quality of data presented.
One of the reasons for the choice of tomato ketchup is
that its life cycle represents a rather common food-pro-
duct life cycle: it includes a harvest, a preservation pro-
cess (seasonal production), storage, transportation and,
finally, further processing into a consumer product. The
conclusions are of course specific for the tomato ketchup
studied; however, similar results could be expected for
jam and juices. Another reason for the choice of ketchup
was the interest of the Swedish ketchup producer and
their willingness to participate. Without such support, it
would not have been possible to obtain site-specific data.
Acknowledgements
The Swedish Waste Research Council is gratefully
acknowledged for funding. We also wish to thank
warmly all who have kindly assisted us with site-specific
data and patiently answered our many questions. Special
thanks to Lora Sharp-McQueen who reviewed the
English of this paper.
References
[1] Leach G. Energy and food production. Guilford: IPC Science and
Technology Press, 1976.
[2] Singh RP, editor. Energy in food processing. Amsterdam: Elsev-
ier, 1986.
[3] Jolliet O. Revue Suisse de Viticulture Arboriculture Horticulture
1993;25:261–7.
[4] Weidema BP, Pedersen R, Drivsholm TS. Life cycle screening
of food products—two examples and some methodological pro-
posals. Denmark: Danish Academy of Technical Sciences,
Lyngby, 1995.
[5] Andersson K, Ohlsson T, Olsson P. In: Ceuterick D, editor. Pre-
prints from International Conference on Application of Life Cycle
Assessment in Agriculture, Food and Non-Food Agro-Industry
and Forestry: Achievements and Prospects, VITO, Mol
(Belgium), 1996:139–57.
[6] Habersatter K, Widmer F. Oekobilanz von packstoffen stand
1990. Schriftenreihe Umwelt Nr. 132. Bern: BUWAL, 1991.
[7] Biondi P, Monarca D. Rivista di ingenieria agraria 1989;3:129–
45.
[8] Tillman A-M. Godstransporter i livscykelanalys. Schablonvärden
för energianvändning och emissioner. Gothenburg: Technical
Environmental Planning, Chalmers University of Technology,
1994.
288 K. Andersson et al. / Journal of Cleaner Production 6 (1998) 277–288
[9] Tillman A-M, Baumann H, Eriksson E, Rydberg T. Packaging
and the environment. Life-cycle analyses of selected packaging
materials. Quantification of environmental loadings. Gothenburg:
Chalmers Industriteknik, 1991.
[10] Eriksson E, Svensson G, Lövgren G, Blinge M, Svingby M,
Ölund G. Transporters miljöpåverkan i ett livscykelperspektiv.
REFORSK report FoU 126. Gothenburg: Chalmers Industritek-
nik, 1995.
[11] Person L, Zackrisson M. Life cycle assessments including the
working environment—a case study of fridge/freezers from Elec-
trolux, Part II: DATA, 95833. Gothenburg: IVF, 1995.
[12] Lindfors L-G, Christiansen K, Hoffman L, Virtanen Y, Juntilla
V, Hanssen O-J, Rønning A, Ekvall T, Finnveden G. Nordic
guidelines on life-cycle assessment. Copenhagen: Nordic Council
of Ministers, 1995.
[13] Tillman A-M, Ekvall T, Baumann H, Rydberg T. Journal of Cle-
aner Production 1994;2:21–9.
[14] IPCC (Intergovernmental Panel on Climatic Change). Radiative
forcing of climate change—the 1994 report of the Scientific
Assessment Group of IPCC, John Houghton. Bracknell: Meteoro-
logical Office, 1994.
[15] Finnveden G, Andersson-Sköld Y, Samuelsson M-O, Zetterberg
L, Lindfors L-G. In: Lindfors L-G, editor. Product life cycle
assessment—principles and methodology. Copenhagen: The
Nordic Council of Ministers, 1992:172–231.
[16] Lindfors L-G, Christiansen K, Hoffman L, Virtanen Y, Juntilla
V, Hanssen O-J, Rønning A, Ekvall T, Finnveden G. LCA-Nordic
technical report no. 10. Copenhagen: Nordic Council of Minis-
ters, 1995.
[17] Andersson-Sköld Y, Grennfelt P, Pleijel K. Journal of the Air
and Waste Management Association 1992;9:1152–8.
[18] Heijungs R, editor. Environmental life cycle assessment of pro-
ducts: guide and backgrounds. Leiden University: Centre of
Environmental Science, 1992.
[19] The Tellus packaging study. Boston: Tellus Institute, 1992.
[20] Peter D. In: Schaltegger S, editor. Life Cycle Assessment
(LCA)—quo vadis? Basel: Birkhäuser Verlag, 1996:95–129.
[21] Wilting H. An energy perspective on economic activities. The
Netherlands: University of Groningen, 1996.
[22] Att äta för en bättre miljö, rapport 4830. Stockholm: Nat-
urvårdsverket, 1997.
[23] Cowell S, Clift R. Life cycle assessment for food production sys-
tems. Proceedings no. 375. Peterborough: The Fertiliser
Society, 1995.
[24] van Zeijts H, Leneman H, Wegener Sleeswijk A. In: Ceuterick
D, editor. Preprints from International Conference on Application
of Life Cycle Assessment in Agriculture, Food and Non-Food
Agro-Industry and Forestry: Achievements and Prospects, VITO,
Mol (Belgium), 1996: 69–76.
[25] Mattsson B. In: Kristensen NH, Høgh-Jensen H, editors. New
research in organic agriculture, down to earth—and further afield.
Proceedings Vol. 2, 11th International Scientific IFOAM Confer-
ence, Copenhagen, 1996:180–4.
[26] Audsley E, Alber S, Clift R, Cowell S, Crettaz P, Gaillard G,
Hausheer J, Jolliet O, Kleijn R, Mortensen B, Pearce D, Roger
E, Teulon H, Weidema B, van Zeijts H. Harmonisation of
environmental life cycle assessment for agriculture. Final report
of concerted action AIR3-CT94-2028. Bedford: Silsoe Research
Institute, 1997.